1st Principle Thinking - Part 1

First Principle Thinking for P4 medicine model

Aristotle defined a first principle as “the first basis from which a thing is known”.

For this I want to break down the problem into fundamental truths and reason up from there. Challenging my assumptions and questioning what I know to develop a deeper picture.

The vision

“Medicine that is predictive, preventative, personalised and participatory” – Leroy Hood

Describing the P4 medicine model

Predictive – Uses data to anticipate disease risk before symptoms appear, such as genetic risk for diabetes.

Preventative – Instead of focusing on treating an illness, the aim is to actively prevent it through early interventions.

Personalised – This is tailoring medical care to an individual’s unique biology, environment, lifestyle and preferences rather than a one size fits all treatment.

Participatory – Empowers patients to actively engage in their own health. A good example is wearables.

How do we measure the health of an individual? Where can we gather the most data for health outcomes? What measurements predict future health outcomes with high accuracy?

In the human body DNA is transcribed into RNA which is translated into proteins (Fig 1.).

Figure 1 - Source: New diagnostic molecular markers and biomarkers in odontogenic tumours DOI: 10.1007/s11033-021-06286-0

Each of these -omics and at each stage contains a wealth of information (Fig 2.).

Omics	Predictive	Preventative	Personalised	Participatory
Genomics
Epigenomics
Transcriptomics
Proteomics
Metabolomics

Figure 2. Green alignment, yellow semi-alignment, red no-alignment as of yet

Though genomics is excellent for identifying risk, how can you track changes. It’s the most personalised of the omics, but it is static and there’s no day to day feedback. Companies like Oxford Nanopore are excellent at shortening the time to screen a human genome and gain genetic and epi-genetic information.

Metabolomics reveals early functional changes, it is comprised of endogenous and exogenous metabolites from the environment, microbiome and lifestyle. It reflects the current state of the individual and has measurable changes through non-invasive means that are measurable (Fig 3.).

Within metabolomics there are many sub fields. Lipidomics, fluxomics, exometabolomics, volatolomics, spatial metabolomics all with their own specific classes of molecules or techniques. There are also different approaches in experimental strategy, whether it is untargeted, targeted or widely targeted metabolomics.

Figure 3. There are 4240 known urine metabolites in the human metabolome database (HMDB), 1424 of which have literature associations to a diverse set of human conditions. a Pie chart based on counts of HMDB chemical taxonomy (Super Class) for all metabolites identified in this study (n = 101). b Pie chart based on counts of each HMDB chemical taxonomy for metabolites all of HMDB. c Treemap of diseases, scaled by number of metabolites with literature associations (source)

There are over 200,000 + compounds in the human metabolome database (HMDB) of which there are 4,240 known urine metabolites, 1424 of which have literature associations to a diverse set of human conditions.

Where can we gather the most data? But we can refine this question to within the P4 framework. What’s the minimum data needed to maximise predictive power for health outcomes?

There are many biofluids: Plasma / Serum, Blood, Saliva, Tears, amniotic fluid, urine, breast milk, colostrum, bronchial lavage, cerebrospinal fluid, peritoneal fluid, pleural fluid, seminal fluid, interstitial fluid.

To rank them in terms of metabolite information such as biomarker diversity, diagnostic richness, physiological representativeness, accessibility: Plasma / Serum, Whole Blood Urine, Cerebrospinal Fluid (CSF), Interstitial Fluid (ISF),  Breath (Exhaled Air), Saliva, Seminal Fluid, Breast Milk / Colostrum,  Tear Fluid, Bronchoalveolar Lavage (BAL), Pleural Fluid, Peritoneal Fluid, Amniotic Fluid.

Let’s only have the non-invasive fluids, meaning no needles for collection: Urine, Breath, Saliva, Seminal Fluid, Breast milk, Tear Fluid

For non-invasive biomarker measuring urine contains the most information, but this doesn't mean it is without challenges or orders of magnitude better than other fluids. See Owlstone Medical article for breath biomarker analysis (link)

What data is there in urine? What is the range of data?

There’s both a human metabolomic data base (HMDB) and a Urine Metabolomic Data Base (UMDB) maintained by the Wishart group.

Urine – Simplistic view with concentrations normalised to that sample’s creatinine concentration

• Water (95%)

• Nitrogenous waste products

  • Urea (2 %) (22.5 ± 4.4 mM/mM creatinine)

  • Creatinine (10.4 ± 2.0 mM)

  • Hippuric Acid (298.5 ± 9276.8 µM/mM creatinine)

  • Ammonia

  • Amino acid derivatives

  • Electrolytes

    • Sodium (14.7 ± 9.0 mM/mM creatinine) , potassium, chloride, calcium, magnesium, Phosphate, Sulfate, bicarbonate, Ammonia (2.8 ± 0.9 mM/mM creatinine)

• Metabolites

  • UMDB ~3,100 identified metabolites (link)

  • Hormones

    • Oxytocin (0.9 ± 0.1 pM/mM creatinine)

    • Angiotensin II (1.2 ± 0.2 µM/mM creatinine)

    • Melatonin (3.3 ± 2.7 µM/mM creatinine)

  • Cells

  • Proteins

  • Organic Acids

  • Sugars

  • Vitamins

• Exogenous compounds from lifestyle, medicines & environment

  • Pharmaceutical drugs

  • Microbial products

  • Environmental pollutants (pesticide residues, plasticiser metabolites)

(Note all concentrations are normalised to creatinine)

Conclusions

The current lower limit of detection for metabolites in urine is in the low pM/mM creatinine range. The concentration of analytes in urine spans nearly 11 orders of magnitude. Average metabolite in urine can vary by ± 50% from person to person with some varying as much as ± 350% (such as normetanephrine (0.00085 ± 0.00317 µM/mM creatinine).  Therefore, drawing conclusions about potential disease biomarkers without properly taking this variation would be ill-advised.

The range of information needed to be collected by an analytical instrument is vast. Molecular weigh range is Urea (60 Da) – Uromodulin tamm-Horsfall Protein (~95-100 kDa) in addition to the concentration differences and variations.

However, the insight from data especially combined with wearables is massive. To put it into perspective, through urine metabolite identification and quantification we can see lifestyle factors like exercise, nutrition, sleep, OTC drug usage. In addition, it also contains metabolite information relating to cancer, pollutants, pesticides. Making a huge picture of an individual's current state.

Unlike blood, urine is non-invasive, depending on method requires no change in the person’s habit. There are also many barriers to giving and receiving blood in African cultures (link) and religious groups like Jehovah’s witnesses.

However, to challenge my assumptions, maybe one day giving blood would essentially be non-invasive thus giving the most comprehensive metabolite data. Ideally a platform that is able to measure different bodily fluids to identify and quantify metabolites is necessary. In addition, when technology improves we will be measuring more than one omics method in depth to give an integrative personal omics profile (iPop) as detailed by Michael Snyder (link).

Sources

The Human Urine Metabolome – doi:10.1371/journal.pone.0073076

Physiological conditions can be reflected in human urine proteome and metabolome – doi:10.1586/14789450.2015.1094380

Real-time health monitoring through urine metabolomics doi:10.1038/s41746-019-0185-y